Statement of Novelty

A new strategy (combining enzymatic treatment and subcritical water extraction) for green extraction of saffron (Crocus sativus) tepals polyphenols was developed.

Introduction

Plant by-products originating from the agri-food industry have been demonstrated to contain large amounts of valuable compounds, including polyphenols that could be recovered and used as biologically active substances or as natural food ingredients, replacing synthetic additives [1].

Saffron, Crocus sativus L., is systematically cultivated in Greece, Iran, Spain, Morocco and other countries. Its dried stigmas (saffron spice) is used in foods both for the intensive flavouring and colouring properties [2]. Since approximately 150,000 flowers are needed to produce 1 kg of saffron spice [3], representing only 7% of the total flower weight [4], a large amount of by-product (mainly tepals) is generated. Saffron tepals are rich in biologically active compounds such as apocarotenoids and their glycosides, monoterpenoids, polyphenols, fatty acids and phytosterols [5]. Saffron tepal polyphenols, particularly flavonol glycosides and anthocyanins, have been demonstrated as potent antioxidants [6, 7].

To overcome the problems encountered when using conventional organic solvent extraction, in particular to reduce the energy consumption and adverse effects on the environment, various green processes have been developed [8]. Among them, subcritical water extraction (SWE), also known as pressurized hot water extraction, is one of the most promising techniques. In this method, water is heated under sufficient pressure to maintain its liquid state at a temperature above the boiling point (100 °C), but without reaching the critical point (374 °C). Under these conditions, water has a lower dielectric constant, thus behaving similarly to less polar organic solvents like methanol and ethanol [9]. SWE has been successfully applied for the recovery of polyphenols from plant by-products such as apple [10] or grape [11, 12] pomace, potato peels [13] and pistachio hulls [14].

Co-pigmentation is a colour-stabilising mechanism in which an anthocyanin (pigment) and another, usually colourless, molecule (co-pigment) form complexes predominately by π–π stacking interactions in the aqueous environment of the plant vacuoles [15]. Mostly with polyphenols acting as co-pigments, this biological principle is assumed to be responsible for the colour variations that occur within epidermal flower tissues in a pH range where isolated anthocyanins are virtually colourless [16]. While the significance of co-pigmentation for the colour evolution during red wine aging has been thoroughly investigated [17, 18], it has not been frequently considered with respect to anthocyanin stability in fruit processing.

Recently, an extensive cell wall degradation by commercial enzymes was demonstrated to improve the extractability of polyphenolic antioxidants from saffron tepals [19]. Therefore, the present study evaluated the combined effect of subcritical water and enzymatic treatment on the yield of polyphenols from saffron tepals. The polyphenolic composition of the by-product extract was characterised using LC–MS analysis. Finally, the co-pigmentation efficiency of saffron tepal polyphenols was assessed in model solutions of strawberry anthocyanins. The retention of strawberry colour has always been a technological challenge due to the low total content of strawberry anthocyanins [20] and their inherent heat and light sensitivity [21].

Materials and Methods

Chemicals

The following standards were used for LC–MS identification: apigenin (Alfa Aesar, Thermo Fisher Scientific, Kandel, Germany); quercetin (Sigma-Aldrich, Steinheim, Germany); kaempferol (Phytoplan, Heidelberg, Germany). For analytical purposes the following reagents were used: Folin–Ciocalteau’s reagent (Merck, Darmstadt, Germany); gallic acid monohydrate (Fluka, Buchs, Switzerland). The adsorbent resin Amberlite® XAD16N was provided by Sigma-Aldrich (Steinheim, Germany). All other reagents and solvents were of analytical or HPLC grade. Deionized water was used throughout.

Plant Materials and Enzyme Preparations

The flowers of Crocus sativus L. were collected in the autumn of 2019 in the region of Kozani (Ptelea village), Greece. After removing the stigmas, the saffron tepals (ST) were dried at room temperature (25–27 °C) for 1 week before final hot air drying (60 °C, 1 h). Dried saffron tepals were milled in a laboratory mill (model TSM6A011W/01, Robert Bosch Hausgeräte GmbH, München, Germany) and sieved (700 μm laboratory sieve). The final saffron tepal powder was stored in plastic bags at room temperature until usage.

Strawberries (Fragaria × ananassa Duch., cultivar ‘Clery’; harvest 2020) were from the local market in Plovdiv and were stored frozen until usage.

The following commercial enzyme preparations were used: cellulolytic preparation Celluclast BG (Novozymes A/S, Bagsvaerd, Denmark); hemicellulolytic preparation Xylanase AN (Biovet JSC, Peshtera, Bulgaria).

Extraction of Saffron Tepal Polyphenols

A static mode subcritical water extraction (SWE) was performed using a homebuilt extraction system (constructed by Department of Industrial Heat Engineering, UFT-Plovdiv, Bulgaria) (Fig. 1). Water was heated to the desired temperature in a vessel 1 by an electric heater (InnoSolv LLC, Plovdiv, Bulgaria). A stainless steel extraction vessel (2), with a working volume of 1.7 L, was made by InnoSolv LLC, Plovdiv, Bulgaria. The extract was collected in a cooling vessel (3). To avoid boiling of the water, a back pressure was maintained in the system using carbon dioxide from gas cylinder 4. The heating and extraction vessels were thermally insulated.

Fig. 1
figure 1

Schematic diagram of the subcritical water extraction system: 1—water heating vessel; 2—extraction vessel; 3—cooling vessel; 4—CO2 gas cylinder

Full size image

Ground saffron tepals (100 g, particle size < 700 μm) were extracted (160 °C, 0.9–1.0 MPa, 15 min) using a solid-to-solvent ratio of 1:11 (w/v). After separating the extract, the solid residue from the first SWE step (SWE-1) was directly mixed with 60 mL of an enzyme solution (3%, w/w), containing an enzyme preparations mixture (cellulase:xylanase = 1:1) in acidified (HCl, pH 4.0) water. The mixture was incubated for 3.5 h at 50 °C. The enzymatically treated residue was transferred directly in the extraction vessel and was subjected to second SWE step (SWE-2; 1:11 solid-to-solvent, 160 °C, 0.9–1.0 MPa, 15 min). The obtained extracts were clarified through a paper filter.

The same procedure was applied for the control sample, but without enzymatic hydrolysis between the SWE steps.

After concentration under vacuum (40 °C), the combined extract originating from the enzyme-assisted SWE process was purified as described by Shikov et al. [22]. The purified extract was fed into a Büchi B-190 mini spray dryer (BÜCHI Labortechnik AG., Flawil, Switzerland). The operating conditions were as follows: aspirator rate—75% (26 m3/h); drying air temperature—136 °C; pump rate—6.5 mL/min; air-flow 600 L/h; outlet temperature − 62 °C. The dried extract was stored in a glass jar at -20 °C until usage.

Extraction of Strawberry Anthocyanins

Frozen strawberries were thawed and manually squeezed in a beaker. The homogenized purée (approximately 3 kg) was extracted overnight at 4 °C using 3 L of methanol acidified with hydrochloric acid (1%, v/v). The extraction mixture was filtered through a paper filter on a Buchner funnel, and the organic solvent was evaporated under vacuum (40 °C). To remove sugars, salts, and amino acids from the crude extract, the sample was purified as described by Shikov et al. [22]. The organic solvent of the eluate was evaporated under vacuum (40 °C). To separate anthocyanins from colorless phenolics, further purification was performed by extracting the aqueous phase three times with the same volume of ethyl acetate. After evaporation and concentration under vacuum (40 °C), the extract was frozen (− 20 °C, overnight) and freeze dried (Cryodos-50, Telstar, Terrassa, Spain) at − 46.6 °C under a pressure below 0.191 kPa for 48 h. The lyophilized extract was ground and stored in a glass jar at − 20 °C until usage.

LC–MS Analysis

Prior to analysis, 50 mg of saffron extract powder was vortexed with 2 mL 70% methanol and sonicated for 10 min. After centrifugation (10 min, 9600×g, 20 °C), the methanol layer was collected in a 5 mL volumetric flask and the extraction procedure was repeated. The combined methanol layers were filled up with water and filtered through a disposable filter (regenerated cellulose, pore size 0.20 μm, Macherey–Nagel GmbH and Co. KG, Düren, Germany). The filtered sample was tenfold diluted with water.

Analysis was performed according to Dilberger et al. [23] with some modifications. Acquity UHPLC system by Waters (Milford, MA, USA) coupled with an LTQ XL linear ion trap by Thermo Scientific (Waltham, MA, USA) was used. The Acquity UHPLC consisted of a binary pump, an autosampler (cooled at 20 °C), a column oven set at 40 °C, and a diode array detector scanning from 190 to 700 nm. An Acquity HSS-T3 RP18 column (150 mm × 2.1 mm; 1.8 μm particle size) combined with a pre-column (Acquity UPLC HSS T3 VanGuard, 100 Å, 2.1 mm × 5 mm, 1.8 μm), both from Waters (Milford, MA, USA), were used for chromatographic separation with water (A) and acetonitrile (B) as eluents, both acidified with formic acid (0.1 + 99.9, v + v). The flow rate was set at 0.5 mL/min with a gradient elution: 0.0–1.5 min: 1.0% B; 1.5–18.0 min: 1.0–25.0%B; 18.0–30.0 min: 25.0–100% B; 30.0–35.0 min: 100% B; 35.0–40.0 min: 100–1% B. The injection volume was 5 μL. A linear ion trap with an electrospray interface operating in negative ion mode was used. The mass spectrometry (MS) settings were as follows: sheath gas flow: 70 arb; auxiliary gas flow: 10 arb; sweep gas flow: 1 arb; capillary temperature: 350 °C. Negative mode: source voltage: 2.50 kV; capillary voltage: − 21.00 V; multipole 00 offset: + 5.50 V; multipole 0 offset: + 6.25 V; multipole 1 offset: + 10.00 V; front lens: 6.75 V. Normalized collision energy: MS2 35%, MS3 35%, using helium as collision gas. The system was controlled using Xcalibur Software 2.2 SP1.48 (Thermo Scientific, Waltham, MA, USA).

Spectrophotometric Methods for Chemical Analysis

All measurements were performed on a Helios Omega UV–Vis spectrophotometer equipped with VISIONlite software (all from Thermo Fisher Scientific, Madison, WI, USA), using 1 cm path length cuvettes.

The content of total polyphenols (TPP) was determined by the method of Singleton and Rossi [24] as modified by Dinkova et al. [25]. The results were presented as equivalents of gallic acid (GAE). Prior to analysis, an aliquot (10 mL) of saffron tepal extract was transferred into a 50 mL volumetric flask using 40 mL acidified (0.1% HCl) methanol. After extraction for 24 h at 10 °C, the flask was filled up to the mark with the solvent and filtered through a paper filter. Extractions were performed in triplicate.

The amount of total monomeric anthocyanins of the strawberry extract was determined by the pH-differential method [26]. The result was calculated using a molar extinction coefficient of 15,600 L/(mol cm) and molecular weight of 433.2 g/mol [27,28,29] and expressed as equivalents of pelargonidin 3-glucoside (PGE).

Co-pigmentation Experiments

The co-pigmentation effect of saffron tepals polyphenols was evaluated at different molar pigment/co-pigment ratios. Stock solutions of strawberry (prepared as described in “Extraction of Strawberry Anthocyanins” section) and saffron tepal extracts (prepared as described in “Extraction of Saffron Tepal Polyphenols” section) were prepared on the basis of total anthocyanins (quantified as PGE) and total polyphenols (quantified as GAE), respectively. The stock solutions were prepared in McIlvaine buffer (0.1 M, pH 3.4) [30]. Model solutions of strawberry anthocyanins (1 × 10–4 M) with and without co-pigment addition were obtained by mixing 5 mL of both stock solutions and were then left for equilibration (30 min at 25 °C). The following molar pigment/co-pigment ratios were used: 1:0; 1:2.5; 1:5; 1:7.5; 1:10. Model solutions were prepared in triplicate.

The thermal stability of the mixtures was evaluated via heating experiments at 85 °C performed in a W3 laboratory water bath (VEB MLW Prüfgeräte-Werk Medingen, Freital, Germany). After 0, 60, 120, 180 and 240 min heating the glass tubes were immediately cooled in an ice bath to stop thermal degradation of anthocyanins. An aliquot of each sample was filtered through a Filtrak 390 paper filter (VEB Spezialpapierfabrik Niederschlag, Niederschlag/Erzgebirge, Germany).

In order to achieve spectrophotometrically measurable absorption values (absorbance < 2 AU), the model solutions were diluted (1:1, v/v) with the McIlvaine buffer. Visible spectra were recorded in a range of 450 to 700 nm. The absorbance at the wavelength of maximum absorption and the wavelength of the absorption maximum were taken down.

The co-pigmentation effect (CE) and spectral shift of the absorption maximum (Δλmax) were determined as follows Gras et al. [31]:

$${\text{CE }}\left( \% \right) \, = \, \left( {\left( {{\text{A }} - {\text{ A}}_{0} } \right)/{\text{A}}_{0} } \right) \, \times { 1}00,$$

where A0 is the absorbance at the wavelength of maximum absorption (λmax) of the anthocyanin solution without co-pigments, and A is the absorbance at λmax of the respective anthocyanin solutions with added co-pigments;

$$\Delta \lambda_{{{\text{max}}}} = \lambda_{{{\text{maxCP}}}} - \lambda_{{\max_{0} }} ,$$

where λmax0 is the wavelength of the absorption maximum of the anthocyanin solutions without co-pigments and λmaxCP is the wavelength of the absorption maximum of the corresponding solutions with added co-pigments.

All measurements were performed with a Helios Omega UV–Vis spectrophotometer equipped with ColorCalc Basic software (all from Thermo Fisher Scientific, Madison, WI, USA), using 1 cm path length cuvettes.

Statistical Analysis

The reported in the present study results are the mean values of at least three analytical determinations and the coefficients of variation, expressed as the percentage ratios between the standard deviations and the mean values. The means were compared using one-way ANOVA, performed with Microsoft Excel, and Tukey’s test at a 95% confidence level.

Results and Discussion

Yield of Saffron Tepal Polyphenols

A novel process, enzyme-assisted subcritical water extraction, was established on a pilot-plant scale (Fig. 2). This process combines two subcritical water extraction (SWE) steps with enzymatic treatment after the first SWE (SWE-1). The second SWE (SWE-2) was also used as the enzyme inactivation step.

Fig. 2
figure 2

Flow diagram of the new process for recovery of polyphenols from saffron tepals: ‒‒‒ main process step; --- additional process step

Full size image

Ahmadian-Kouchaksaraie et al. [32] determined optimum conditions (159 °C, 54 min) of SWE for recovery of total polyphenols from saffron tepals. However, the efficient extraction of flavonols, which are the main phenolic compounds in saffron tepals [33], has been shown to be faster (15–30 min) at extraction temperatures of 145–165 °C [34, 35]. Therefore, the SWE in the present study was performed at extraction temperature of 160 °C and total extraction time of 30 min (2 × 15 min steps).

The crude extract was purified using adsorber technology [36], which has been successfully applied to the recovery of phenolic compounds from plant by-products such as apple [37] and grape [38] pomace.

The enzymatic treatment resulted in an approximately fourfold increase in the polyphenolic yield (up to 8.37 g GAE/kg saffron tepals for the total SWE) compared with the control sample (without enzymatic treatment) (Fig. 3).

Fig. 3
figure 3

Yields of total polyphenols (g GAE/kg saffron tepals) depending on the different extraction processes—with or without enzymatic treatment between the two subcritical water extraction (SWE) steps

Full size image

This result confirms the efficiency of the enzymes applied in the cell wall degradation. The structure of the plant cell walls illustrates the mechanism of synergistic effects between xylanase and cellulase. In the cell walls, xylan chains (arabinoxylans) are thought to bind to cellulose microfiber surfaces and bind several cellulose microfibers to form crosslinked structures [39]. This cell wall structure model suggests that the degradation of xylan networks between cellulose microfibers by xylanases can allow cellulases to access and degrade the cellulose. Also, the degradation of cellulose microfibers can help xylanase to access and further degrade xylan chains. This degradation model could explain the unique characteristics of synergy between xylanase and cellulase [40].

Therefore, the mechanism proposed above, in combination with the high temperature and pressure applied during the SWE, may be a plausible explanation for the enhanced extractability of polyphenols from saffron tepals.

Characterization of Saffron Tepal Polyphenols

The identification of flavonoids was based on the UV spectra, MSn spectra and chromatographic behaviour, and by comparison with previous studies and to reference compounds where available.

The UHPLC-DAD chromatogram of the saffron tepal extract at 350 nm is shown in Fig. 4. Retention times, UV maxima, and MSn data are shown in Table 1.

Fig. 4
figure 4

UHPLC-DAD chromatogram at 350 nm of Crocus sativus tepal extract. For peak assignment see Table 1

Full size image
Table 1 Characteristics of phenolic compounds in saffron tepal extract
Full size table

Compound 1 exhibited a molecular mass [M−H]− at m/z 787. Diagnostic fragments were recorded at m/z 625 [M−H−162]−, 300 [M−H−162−325]−, 445 [M−H−162−180]− and 301 [M−H−162−324]−. The fragmentation behavior is in accordance with literature data and corresponds in particular to the loss of glucose in position 7 (m/z 625) and of the terminal glucose of the sophoroside moiety (m/z 445) [41, 42]. Homo- and heterolytic cleavage of the glycosidic bond (m/z 300 and m/z 301) was observed in the MS3 experiment of m/z 625, releasing the aglycone quercetin. Thus, this compound was tentatively characterized as quercetin 3-O-sophoroside-7-O-glucoside. Analogously, peak 2 (m/z 771), producing in the MS3 experiment the characteristic fragment of the aglycone kaempferol at m/z 285, was attributed to kaempferol 3-O-sophoroside-7-O-glucoside [41].

The 3,7-diglucosides and 3-sophorosides of flavonols have the same m/z value. Differentiation of both was still possible because of four diagnostic fragments produced by the loss of 324 amu, 180 amu, 162 amu, and 120 amu. In 3,7-diglucosides, a main fragment ion [M−H−162]− can be observed in the MS2 experiment with concurrent absence of the product ion [M−H−180]− [42]. In both the 3,7-diglucosides and the 3-sophorosides, the loss of 324 amu is specific, but in 3-sophorosides it represents the main fragment. Compound 3 (m/z 609) released the product ions at m/z 447 [M−H−162]−, 489 [M−H−120]− and 285 [M−H−324]−, indicating the typical fragmentation pattern of the 3,7-di-O-glucoside of kaempferol [42]. In contrast, compound 5 (m/z 639) yielded the main fragment ions at m/z 315 [M−H−324]− and 477 [M−H−162]−. In the extracted ion chromatogram of m/z 639, a second peak was detected referring to the isorhamnetin 3-sophoroside due to typical product ions at m/z 315 [M−H−324]−, 459 [M−H−180]−, 519 [M−H−120]− and 477 [M−H−162]− (data not shown) [41, 42]. Therefore, compound 5 was assigned to 3,7-di-O-glucoside of isorhamnetin, even though the main product ion was produced by the simultaneous loss of both sugar moieties. The fragmentation of compounds 4 and 6 is typical for a 3-O-sophoroside, characterized by the loss of 324 amu during the MS2 experiment, which released the aglycone. The loss of the terminal glucose from the sophorosyl moiety 180 amu [41, 42] was also observed. Therefore, the 3-O-sophoroside derivatives of quercetin (4) and kaempferol (6) were identified. Consistent with Goupy et al. [43], kaempferol 3-O-sophoroside was the major phenolic compound in Crocus sativus tepals.

After MS2 fragmentation, peaks 7 (m/z 593) and 9 (m/z 623) exhibited product ions at m/z 285, 447 and m/z 315 and 477, respectively. These diagnostic fragments are indicative of a rutinosyl moiety, which allowed the identification of compound 7 as kaempferol 3-O-rutinoside and compound 9 as isorhamnetin 3-O-rutinoside [6, 43].

In accordance with literature data, the 3-glucosides of quercetin (8), kaempferol (11) and isorhamnetin (12) were also characterized by the loss of the glucose moiety [M−H−162]−, by their UV spectra and by their elution order in comparison to the other flavonol glycosides [6, 42, 43].

Two acylated flavonol glycosides were tentatively identified by their fragmentation in MS2 and MS3 experiments. For compounds 10 and 13, product ions at [M−H−324−42]− and [M−H−162−42]−, corresponding to acetyldiglycosyl and acetylglycosyl units, respectively, were detected. Therefore, these compounds were characterized as kaempferol 3-O-(6''-acetyl)-glycoside-7-O-glycoside (10) and kaempferol 3-O-(6''-acetyl)-glycoside (13).

Peaks 14 and 16 were attributed to the aglycones quercetin and kaempferol, respectively. Interestingly, compound 15 produced a molecular ion at m/z 269 and fragment ions m/z 225, 149 and 201, which are characteristic of the flavone apigenin [44,45,46]. The identification of these three compounds was confirmed by comparison with authentic reference compounds. To our knowledge, this is the first report on presence of apigenin in Crocus sativus tepals, however in trace amounts. Carryover during chromatography can be ruled out by the analysis of blank samples.

Co-pigmentation Efficiency of Saffron Tepal Polyphenols in Model Solutions of Strawberry Anthocyanins

The interactions between strawberry anthocyanins (pigment) and the saffron tepal polyphenols (co-pigment) were studied spectrophotometrically in model solutions at pH 3.4, which is similar to that of strawberry juice. The kinetic reaction approach was used which allowed studying the influence of saffron tepals polyphenols on the thermal stability of strawberry anthocyanins.

To investigate the co-pigmentation efficiency against strawberry anthocyanins, the saffron tepal extract was purified using a polymeric adsorber resin, which resulted in 6.1% content of total polyphenols (quantified as GAE).

As shown in Fig. 5, in accordance with Petrova et al. [47], the magnitude of the co-pigmentation effect (CE) increased with the co-pigment concentration. For the solutions with the highest molar pigment/co-pigment ratio (1:10), the CE amounted was 116%.

Fig. 5
figure 5

Visible absorption spectra (450–650 nm) of strawberry anthocyanins in model solutions with added saffron tepal polyphenols. Different molar pigment/co-pigment ratios: (a) 1:0; (b) 1:2.5; (c) 1:5; (d) 1:7.5; (e) 1:10

Full size image

However, in contrast to the addition of rose petal polyphenols [47], no bathochromic shifts (Δλmax) were observed for the saffron tepal polyphenols, implying that the co-pigmentation reaction does not change the colour tonality of the strawberry anthocyanins.

Besides spectral variations, some thermodynamic parameters of the co-pigmentation reaction were determined using the following equation:

$${\text{ln}}\left[ {\left( {A - A_{0} )/A_{0} } \right)} \right]\, = \,{\text{ln}}\left[ K \right]\, + \,n\, \times \,{\text{ln}}\left[ C \right],$$

where A is the absorption maximum value for the co-pigmented anthocyanins, A0 is the absorption maximum value for the anthocyanins alone, C is the molar co-pigment concentration, K is the equilibrium constant of the reaction and n is the stoichiometric ratio of the compounds in the complex [48].

Figure 6 shows co-pigmentation effect (ln[(A−A0)/A0)]) as a function of the co-pigment concentration (ln[C]) at time 0. The concentrations of saffron tepals polyphenol used to plot Fig. 6 were 2.5 × 10–4; 5.0 × 10–4; 7.5 × 10–4 and 10.0 × 10–4 M. The plot shows linear relationships with slope close to unity, suggesting a 1:1 pigment/co-pigment complex, which is in agreement with other studies [48, 49]. The calculated value of the equilibrium constant, reflecting stability of the complex formed, was ca. 400 M−1.

Fig. 6
figure 6

Plots of co-pigmentation effect (ln[(A-A0)/A0)]) as a function of the co-pigment concentration (ln[C]), y = 0.835x + 5.926 (R2 = 0.99)

Full size image

Further, pigment degradation was assessed by the absorbance measurements [50]. Half-life (T1/2) values were determined as described by Sadilova et al. [51] which allowed comparing the times for 50% reduction of the anthocyanins in the model solutions with different co-pigment concentrations. Higher half-life (T1/2) means increased anthocyanins stability.

As shown in Table 2, with the increasing of the heating times the absorbance values (Amax) were significantly lowered. As indicated by the high values of the correlation coefficient (R = 0.90–0.99), anthocyanin degradation ideally followed first-order reaction kinetics (Table 2). These findings are in accordance with previous studies reporting first-order degradation kinetics for strawberry anthocyanins both in model systems [22, 52] and real food matrices [53, 54].

Table 2 Absorbance at absorption maximum (Amax) and half-life (T1/2) values of strawberry anthocyanins (pigment) upon heating at 85 °C in model solutions with added saffron tepal polyphenols (co-pigment)
Full size table

The half-life (T1/2) values increased (39–105%) due to the addition of saffron tepal extract, reflecting lower pigment degradation rates in the presence of polyphenolic co-pigments, which is in agreement with other studies [22, 55]. Maximum heat stability was reached at the highest pigment/co-pigment ratio (1:10).

Conclusion

A new process for recovery of saffron tepal polyphenols was developed on pilot-plant scale. Combining 2-step subcritical water extraction and enzymatic hydrolysis resulted in a notably higher recovery rate compared to the control (without enzymatic treatment) sample. Together with the application of adsorber technology for the concentration of phenolic compounds, this new process may provide an innovative approach for valorization of the saffron by-product as a source of natural co-pigments.